Automated network feature weighting-based anomaly detection

We propose in this paper an automated feature weighting method based on fuzzy subspace approach to assign a weight to each network feature depending on its degree of importance in anomaly detection. Fuzzy c-means and fuzzy entropy modeling are used to calculate weight values and k-means vector quantization is used to model network patterns. The proposed method not only increases the detection rate but also reduces false alarm rate as shown in our experiments.